Lending Test Automation Case Study

Customer: Big 4 Australian bank
Industry: Banking and finance
Project: End-to-end lending regression automation

The challenge

125 testers.
Four releases a year.

The bank's lending group relied on a traditional offshore regression model of around 125 people. Regression across its lending landscape took so long that changes could only reach production quarterly, while the business wanted to release much faster and more often.

Lending processes spanned more than 30 interconnected systems, including mainframe, web, client-server, mobile and external credit services. A single loan could pass through many of them, so regression had to follow complete end-to-end processes rather than individual screens or isolated systems.

Scaling the manual and coded approach further would have added cost without shortening cycles. The bank needed a way to run full regression far more often, with fewer people, while keeping confidence in every release that reached its lending customers.

Lending programme team reviewing process charts

What made it hard

More than 30 connected systems: Lending processes crossed mainframe, web, client-server, mobile and external credit services, so every journey involved many hand-offs.
Quarterly release ceiling: Regression effort limited production releases to four a year, far slower than the business and its customers needed.
A large offshore model: Around 125 offshore resources were needed for each regression cycle, adding cost, coordination effort and delay to every release.

Our solution

End-to-end automation
with virtual testers.

PinnacleQM automated the lending group's end-to-end processes in Enginuity, provided as a Platform-as-a-Service, and replaced the offshore regression model with a local team of 12 supported by scalable virtual tester bots running around the clock.

Natural-language scripts followed complete lending processes across all the connected systems, from mainframe to mobile and external credit services. Because the scripts were not coded to each technology, one test could cover a whole loan journey from application to settlement.

Virtual tester bots executed the regression at scale on schedule, enabling daily, weekly and monthly test cycles, rapid defect retesting, and progression and regression testing into weekly production releases, with results integrated into the bank's CI/CD pipeline for every build.

How it was delivered

From automation build
to continuous releases.

Delivery moved the bank from a manual regression model to continuous testing.

01
Design
- Processes mapped
  End-to-end lending processes across more than 30 systems were identified and prioritised for automation by business criticality.
- Platform provided
  Enginuity was provided as a Platform-as-a-Service, so no toolchain had to be built, procured or maintained by the bank.
02
Execution
- Natural-language scripts
  Lending journeys were automated with reusable natural-language scripts spanning mainframe, web, client-server, mobile and external services.
- Virtual testers
  Scalable virtual tester bots executed regression daily, weekly and monthly as required, without extra people at the keyboard.
03
Verification
- Rapid retesting
  Defect fixes were retested quickly by virtual testers, often the same day, keeping weekly releases moving on schedule.
- Weekly production releases
  Progression and regression testing fed weekly production releases through the bank's CI/CD pipeline, with evidence for each run.